Originally Posted By: artie505
I think I"m beginning to understand this.

After looking through your report and my two (Damn design copyrights!), I"m left with these questions:
  1. I assume it's manufacturer option, but do you see any logic as to why some attributes are identified as "Pre-Fail" on one drive and "Life-Span/Old Age" on another?
  2. "Any positive number" in which column of your report?
  3. Would you please explain "If a pre-fail attribute racks up half of the threshold value" graphically? I don't follow it from what I've been looking at.

  1. Why the difference in interpretation?
    1. There can be honest differences of engineering opinion on whether a given attribute is a reliable pre-fail indicator or merely old age.
    2. Although the attributes number and name are defined by the SMART standard the "tolerance" and what its significance is, is up to the manufacturer's engineers as "tweaked" by their marketing department.
    3. You also have to consider those engineer's first or only language may well not be English so the connotation of the significance title could well come into play here as well.
  2. I didn't think that one out too well I probably should have said any positive raw value in a pre-fail indicator but I might not want to be confined to the pre-fail values. (If I had the resources of Google Labs or access to Apple's Data Centers it would be interesting to statistically test and find the optimum set of SMART attributes to choose for SATA HDs and SSDs as well as NVMe drives)
  3. Pick an attribute, say attribute number 1. "Raw Read Error Rate" with a [max] value of 100% and a threshold of 62%. That implies that when the Raw Read Error Rate reaches 62% of the attempts the drive has failed that attribute. (Note I got those numbers from an HGST drive and a raw read error rate that high would indicate the drive's performance was in the toilet.) But back to my suggestion if the Raw Read Error Rate reaches 31% I would want to know it because
    1. the drive performance is suffering
    2. it is likely caused by magnetic media flaking off the platter
    3. Once that starts it is likely to continue and accelerate
    4. I would want to look at the other pre-fail attributes to see if they confirm the Raw Read Error Rate. — The drive may well have already failed but not reached the manufacturer's too rosy failure settings.


Originally Posted By: artie505
And in a different vein, doesn't it seem odd that only one SMART attribute is reported for freelance's SSD?

Remember freelance's drive is a two drive RAID array so it is quite reasonable that even with the hack only one value can be seen and extracted. Reconfiguring the enclosure to another RAID level or if possible turn off RAID so the drives can be used individually and the results might be quite different.

If we knew what it was we were doing, it wouldn't be called research, would it?

— Albert Einstein